The paper examines the changes in the profile of investors on the Warsaw Stock Exchange before and during the COVID-19 pandemic. Our goal is to identify the impact of the pandemic outbreak on portfolios, the intensity of investors’ biases and possible consequences to the decision-making process, manifested in prospect theory.
The research presented in this paper is based on the representative surveys conducted among Polish individual investors in 2019 and 2020. In the first part of the study, we present descriptive statistics of the sample and test for its representativeness. Later, we test the differences between the surveys. In the second part, which aims to identify the investors’ decision-making process influenced by the pandemic outbreak, we use the modified Lucas utility function to identify investors’ biases and apply it to the prospect theory value function.
We find evidence that the market situation observed in 2020 has significantly changed attitudes towards portfolio management, altered investors’ utility functions and shaped their expectations. Our research shows that risk aversion increased while loss aversion decreased during the pandemic. Taken together, these two factors provide a good illustration of the impact of the March 2020 market crash on investors’ preferences – an increase in risk aversion leads to faster closing of loss-making and profitable positions, but the decrease in loss aversion means that investors would be less interested in holding loss-making positions longer than profitable ones. Thus, our study contributes to the behavioral finance literature under the specific conditions of the COVID-19 pandemic.
Our research is based on unique data collected from investors in our online survey just before and during the outbreak of the Covid-19 pandemic. This gives us a valuable opportunity to analyze the changes to portfolios and behavioral biases among investors resulting from a shift in volatility.
Introduction
The significant increase in the share of individual investors in the total turnover of the Warsaw Stock Exchange (WSE) observed during the COVID-19 pandemic prompted us to analyze the profile of investors, their habits and preferences regarding investment decisions. According to the WSE data (2019), the share of retail investors in the Main Market in 2019 was 12%, the same as in the previous year. However, if we look at the NewConnect market, this share was 63% in 2019, that is 3% points higher than the previous year. Looking further into the 2020 data, individual investors accounted for 25% of turnover on the Main Market and 92% on NewConnect. Over the same period, the analysis of market volatility shows that 2019 was relatively calm on the WSE, while 2020 was the year of a huge drop and a sharp rebound, as well as relatively high uncertainty about the future shape of the market (Figure 1). The analysis of volatility on the WSE conducted using MS GARCH models with two regimes proves high volatility periods during the pandemic outbreak which may have an impact on investor behavior. Such sharp and unexpected changes in volatility can have a significant impact on investors’ decisions on the market.
The horizontal axis is labeled with years from 2000 to 2020 in increments of 5 years. The vertical axis is labeled with values ranging from negative 0.10 to 0.05 in increments of 0.05 units. The graph shows four series. A legend box in the upper left area identifies the four series: “ln R W I G” in red, “1-step prediction” in green, “Fitted” in blue, and “Regime 1” in grey. The red vertical spikes representing “ln R W I G,” distributed around 0 with extreme fluctuations visible around 2008 to 2010 and again near 2020. A solid green line close to zero represents “1-step prediction.” A blue line labeled “Fitted” overlaps the prediction line. Grey shaded vertical bands mark high-volatility regimes, labeled “Regime 1.”MS GARCH models indicating high volatility periods (grey background). Source: Authors’ own elaboration
The horizontal axis is labeled with years from 2000 to 2020 in increments of 5 years. The vertical axis is labeled with values ranging from negative 0.10 to 0.05 in increments of 0.05 units. The graph shows four series. A legend box in the upper left area identifies the four series: “ln R W I G” in red, “1-step prediction” in green, “Fitted” in blue, and “Regime 1” in grey. The red vertical spikes representing “ln R W I G,” distributed around 0 with extreme fluctuations visible around 2008 to 2010 and again near 2020. A solid green line close to zero represents “1-step prediction.” A blue line labeled “Fitted” overlaps the prediction line. Grey shaded vertical bands mark high-volatility regimes, labeled “Regime 1.”MS GARCH models indicating high volatility periods (grey background). Source: Authors’ own elaboration
In this article, we aimed to identify investor characteristics and behavioral biases and their changes during the COVID-19 pandemic. The literature recognizes the impact of behavioral biases on individual and institutional investors’ decisions and portfolio performance. Portfolio composition began with Markowitz’s works (1952) and was later the subject of analysis during the pandemic period for optimal portfolio weights, hedge ratios (Belhassine & Karamat, 2021), portfolio rebalancing (Ritika, Mushir, & Suryavanshi, 2021) and left-tail asymmetric dependence (Azimil, 2020), while being inherently specific. Moreover, Merkle (2020) investigated the heuristics summarized by Kahneman and Tversky in prospect theory (1979) according to the COVID-19 impact on expected and experienced returns, with loss aversion reduced by half. Walasek, Mullet, and Stewart (2024) also demonstrated the reduction in the loss aversion parameter captured across risky prospects. Blake, Cannon, and Wright (2021) also analyzed the intensity of behavioral biases according to the sample structure and Mrkva, Johnson, Gächter, and Herrmann (2019) revealed significant correlations with gender, age, financial literacy, as well as income, savings and decision object specifics. Scholars also found that the perception of the investment decision to be made also played a great role in the loss aversion factor (Shang, Duan, & Lu, 2021), which may be an important finding for the sharp change in volatility in our sample. Moreover, Almansour (2023) states that risk perception is significant in the decision-making process, while risk perception stems from biases such as herding and disposition effects.
Noteworthy, a relatively high impact of individual investors on certain markets can lead to a slightly more frequent and stronger occurrence of stock market anomalies, as shown by the research of Łukowski, Gemra, Maruszewski, Śliwiński, and Zygmanowski (2020). We aimed to examine how the COVID-19 pandemic affects the investment decisions and preferences of individual investors, especially in the context of portfolio construction and behavioral aspects of decision-making. The hypothesis to be tested is that the COVID-19 pandemic significantly changed the behavior of investors in the financial markets.
We obtained the following results. First, we created a profile of individual investors based on the surveys we conducted in 2019 and 2020, which were the second largest in Poland, apart from the National Investor Survey (OBI) conducted by the Association of Individual Investors. The structure of the sample allowed us to determine the impact of the pandemic on portfolios, compare it with the OBI survey and test for the sample representativeness. Thus, the data we obtained allowed us to draw statistically significant and representative conclusions in the latter parts of the article.
Second, we described the decision-making process regarding the pandemic outbreak and identified investors’ behavioral biases. Based on the given characteristics and multiple-choice questions in our survey, we could draw conclusions about investors’ risk aversion, loss aversion and hyperbolic discounting. In particular, we could compare their values during a relatively calm period (2019) and market turbulence due to the COVID-19 pandemic (2020). The shift in risk aversion observed in the study, together with the decline in loss aversion in 2020, demonstrates investors’ willingness to close losing positions more quickly than prospect theory suggests and to maintain more diversified portfolios. In turn, this leads them to behave more rationally than the behavioral finance theory would suggest.
Third, our main result is the identification of the change in investors’ utility in 2019 and 2020, which adapts the prospect theory to the case of extreme volatility, as seen during the pandemic outbreak. Three biases identified in the second step allowed us to approximate the changes in investors’ utility during the turbulent period, which in turn suggests that the shift in market volatility made investors more concerned about the risk in their portfolios (as they became more risk-averse) and more rational in the face of losses (as a consequence of the inversion of the utility function in its lossy part – 3rd quarter of the coordinate plane).
Our research fills a gap in the financial literature by linking a rather extreme market situation with an intensification of investors’ behavioral biases. First, we add to the discussion on the portfolio shape of investors during the pandemic outbreak as mentioned by Omura, Roca, and Nakai (2021), Yoshino, Taghizadeh-Hesary, and Otsuka (2021), and Belhassine and Karamat (2021). Second, our analysis of investor behavior and the patterns identified can be part of the broader discussion of the impact of COVID-19 on investment decisions (Ortmann, 2020; Zielonka, 2020; Smales, 2021; Talwar, Talwar, Kaur, Tripathy, & Dhir, 2021; Blake et al., 2021; Shang et al., 2021; Almansour, Elkrghli, & Almansour, 2023) and particular biases (Mrkva et al., 2019; Merkle et al., 2020; Gupta & Shrivastava, 2022; Walasek et al., 2024). Finally, our discussion of the investor utility function fits into the debate started by Kahneman and Tversky (1979), continued by Shefrin (2010a) and contradicted by Ruggeri et al. (2020). In the latter case, the results we present are comparable according to the loss aversion factor and the form of the investor’s utility function.
We based the article’s structure on three main parts. After the introduction, we present literature research on the profile of investors on the WSE, as well as a review of behavioral biases and prospect theory, especially during the COVID-19 pandemic outbreak. The second part describes the characteristics of the investors participating in the survey and the methods used in the research. The last – main – part of the paper deals with the analysis of correlations and other relationships from our study, as well as the impact of the pandemic on investors’ preferences. In this section, we also consider the diversification and rebalancing of investors’ portfolios relative to theoretical norms and the changes caused by the pandemic.
Theoretical basis
Given the article’s objective, a review of the literature that discusses the behavior of investors and the way they construct their investment portfolios would be a good benchmark for our research. In the behavioral paradigm, the investor is subject to cognitive-neural imperfections and social influence. Slovic (1972) conducted a psychological analysis of investment decisions. He studied the investor’s response to threats. Shefrin and Statman (2000) paid particular attention to two key emotions, fear and greed, which influence investors’ decisions. On the other hand, neuroeconomics focuses on analyzing the processes that take place in the human brain in the context of economic decision-making (e.g. Glimcher & Fehr, 2013). The literature provides many examples of heuristics, such as mental accounting and loss aversion (Benartzi & Thaler, 1995; Pavlov, 2002), the influence of investors’ experience (Rietz, 1988) and expectations (Fox & Tversky, 1995), availability bias (Benartzi & Thaler, 1995) and trading history (Barberis, Huang, & Santos, 2001). Shefrin (2010b) found that cognitive errors and the resulting heuristics are often persistent and systematic, distort market participants’ beliefs and can change over time. In the prospect theory of Kahneman and Tversky (1979) and Shefrin and Statman (1985), both of which we can consider seminal, the authors summarize heuristics to some extent. They provided many interesting conclusions about the formation of preferences using psychological factors. This led to the emergence of the prospect theory and the disposition effect, which dominated the direction of investor preference research for many years (Kahneman & Tversky, 1992; De Winne, 2021).
More recent literature has also explored the impact of COVID-19 on investors’ behavioral biases, finding that investors’ sensitivity to negative expected returns is twice that of positive returns, and almost halved for experienced returns (Merkle, 2020). Furthermore, Shang et al. (2021) distinguished gambling from investing as a potentially significant factor influencing loss aversion. This could mean that entering a period of high volatility, and therefore limited ability to predict returns, could further increase the loss aversion factor. Blake et al. (2021) offered a different perspective on the factors shaping loss aversion, namely the investor’s characteristics.
Behavioral finance research also provides information on the investor profile. In the context of risk aversion, the tools used by investors and decision-making under uncertainty, Zielonka (2011) provides interesting conclusions. Dacey and Zielonka based their model (2008) on the investor’s investment decisions depending on the identified critical probability. This is a direct reference to the disposition effect (Shefrin & Statman, 1985), which results from the shape of the utility function in prospect theory (Kahneman & Tversky, 1979).
We may find other interesting conclusions on investor behavior in the work of Szyszka (2009), both in the context of tools and behavioral biases. Czerwonka and Gorlewski (2012) presented a description of the behavioral aspects of individual investor decision-making based on a survey. However, the vast majority of research on the profile of an individual investor in the Polish capital market is based on the annual National Investor Survey (Osińska, Pietrzak, & Żurek, 2011; Czerwonka & Oleśniewicz, 2013). This is a relatively broad survey that aims to capture the investors’ profile based on their age, wealth, education, portfolio composition and tools used in trading while missing the whole part of emotions in investing.
Research on portfolio construction includes a 1962 study that found that US investors held an average of 3.41 stocks (Blume & Friend, 1975). At the same time, the correlation between returns in the United States in the 1960s was 0.28 (Campbell, Lettau, Malkiel, & Xu, 2001). The study repeated for the 1980s and 1990s showed a reduction in the number of companies in the portfolios to two and in 2001 to three (Polkovinchenko, 2005) and then to four companies in the 1990s according to the research of Goetzmann and Kumar (2008). Simultaneously, the correlation in the US market fell to 0.08 in the late 1990s (Campbell et al., 2001). Statman’s research (1987, 2004) showed that a well-diversified portfolio should contain about 30 different assets, while marginal effects can be observed for up to 300 companies in the portfolio. For comparison, the work of Gluzicka (2016) helps to identify a portfolio of 30–40 assets on the WSE as a well-diversified portfolio. An interesting finding offered by Belhassine and Karamat (2021) on optimal portfolios in the pre- and post-COVID periods in Chinese and major commodity and financial markets revealed significant changes in optimal portfolio weights (significant decrease for equities and commodities, except for bitcoin), correlations and, consequently, hedge ratios.
There are also several articles on the impact of COVID-19 on investors’ trading behavior. Kizys, Tzouvanas, and Donadelli (2020) studied herding among investors in 72 countries during the coronavirus outbreak. Fernandez-Perez, Gilbert, Indriawan, and Nguyen (2020) focused on the culture effect during the pandemic, demonstrating higher volatility and larger declines in countries with higher risk aversion and lower individualism and comparing it to the situation in 2013 due to SARS. Reis and Pinho (2020) examined the impact of the pandemic on investor rationality. Allam, Abdelrhim, and Mohamed (2020) analyzed the sensitivity of individual and institutional investors to the COVID-19 pandemic in the Egyptian stock market. Corbet, Hou, Hu, Oxley, and Xu (2020) and Qian, Jiang, and Long (2023a, b) analyzed volatility spillovers from Chinese financial markets. Almansour et al. (2023) identified the relationship between risk perception and the decision-making process, where the former is mainly the result of three biases, namely herding, disposition effect and blue-chip bias.
The main conclusion of the pandemic-related articles is the demonstration of behavioral biases such as herding among investors during the first quarter of 2020, which can be mitigated by government responses and short-selling restrictions (Kizys et al., 2020). Scholars also observed higher volatility and larger stock market declines during the first three weeks of the COVID-19 outbreak, which Fernandez-Perez et al. (2020) attribute to the culture effect, another bias affecting the relatively more risk-averse investors. Corbet et al. (2020) analyzed the directional spillovers related to the impact of COVID-19 on Chinese financial markets on cryptocurrency markets, weakening it as a safe haven asset. We may explain this by investors’ need for capital to cover margin calls during the market crash. Conclusions from the Egyptian Stock Exchange cover the sensitivity of individual and institutional investors’ behavior to pandemic-related data, such as daily deaths and total deaths, proving the higher value of trading averages (Allam et al., 2020). Finally, Reis and Pinho (2020) used text media sentiment to compare and analyze the rationality of investors in the US and European markets. They found that US investors tended to be more sensitive to irrational factors, while European investors reacted more rationally to the verified data on the pandemic and the market situation, especially in most of the COVID-19-affected countries. Bauer and Hospodka (2020) and Almeida (2021) also analyzed the importance of data for investment decisions. On the other hand, the case of companies’ profitability and ability to cope with the COVID-19 pandemic is also an important factor in investors’ decision-making, as mentioned by Podedworna-Tarnowska (2023) and Szeiner, Juhasz, Hevesi, and Poor (2023).
The literature on pandemic-related analyses shows that COVID-19 can have a relatively widespread effect on investor behavior. As a result, their investment decisions may be more biased and they may act more irrationally when exposed to such a market situation, but this is not a rule. Furthermore, scholars have observed investor sensitivity to pandemic-related data, which may have implications for portfolio rebalancing and management.
Data and methods
Data-collection and sample characteristics
We based the investor’s profile presented in this article on the responses to anonymous online surveys conducted in July–September 2019 and July–September 2020. The questionnaires (Appendix 1), 31 questions each, were available on the strefainwestorow.pl, and 841 and 767 investors completed it in 2019 and 2020, respectively. Therefore, it was the second largest investor survey on the WSE after the National Investor Survey (2018), which involved 3,912 people. Some of the biases we tested for appeared more than once in the questionnaire to confirm previously answered questions, which allowed us to check the dataset’s reliability. In our opinion, there is a problem with researching individual investors in the Polish capital market because there is no data available on the population. The only data come from the National Depository for Securities, but it refers to the number of brokerage accounts, and one investor can have several accounts. We can only estimate the population of individual investors, including those we consider active, that is those who have made at least one transaction in the year. According to KDPW data, there are 1.4 million brokerage accounts on the WSE (KDPW, 2020), of which 70–80 thousand are active (SII, 2019). Concerning the number of investors’ brokerage accounts, an acceptable error of our findings between 3 and 4% is allowed with a 95% confidence level. Moreover, the comparison of the survey respondents’ structure according to gender, age, portfolio size and composition, as well as investment experience revealed that our sample was consistent with the National Investor Survey based on the fraction test at the 5% significance level. Therefore, our sample was representative.
The survey covered questions from seven areas corresponding to the investor’s profile, allowed for the assessment of the discounting effect over time, biases, information processing, risk aversion as well as anomalies and justified the volatility on the market. Compared to the SII survey, our attempt enabled us not only to draw the profile of the investors but also to reveal the biases behind their decisions and, in a broader context, the behavioral profile of investors on WSE and its changes during the pandemic outbreak. We designed some of the questions as ranges, which on the one hand sacrifices the detail that comes from distribution function analysis, but on the other hand, is an appropriate method for our purposes and allowed us to collect more data as a result of a simpler survey design.
The investors participating in the 2019 survey were predominantly men (93%), with an average age of 44 years. The average age of the women in our study was 48 years. The average age of the entire group was therefore less than 45 years old. Figure 2 (Panel A) presents the age of investors in the 2019 survey. In the 2020 survey, the average age of respondents was 44 years, while for women it was almost 45 years, and for men less than 44 years. We tested the results for the 2019 and 2020 surveys for statistical significance using t-tests and are shown as such further in the text.
The figure shows six panels arranged in a vertical series with two charts per row, the left chart for 2019 and the right chart for 2020. The first panel is labeled “Panel A: Age of the investors participating in the survey in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Investors age” and ranges from 10 to 100 in increments of 10 units. The vertical axis is labeled “Number of observations” and ranges from 0 to 140 in increments of 20 units. The 2019 (left) graph shows a histogram with a low count around ages 10 to 20, rising sharply from 30 to about 100 observations at age 30. The distribution peaks at around 110 observations between ages 35 to 40, remains close to 100 observations for ages 40 to 50, then declines to about 70 observations at age 55. Counts drop to about 40 around age 60, then rise again to 60 to 70 observations between ages 65 to 70, and decline gradually to near zero by age 90. The 2020 (right) graph shows a similar histogram, with fewer than 20 observations around ages 10 to 20. The count increases steeply to around 70 observations at age 30, peaking at about 120 observations between ages 35 to 40, and remaining above 100 through ages 40 to 45. Counts gradually decrease to about 90 at age 50, 60 at age 55, and 30 to 40 between ages 60 to 70, before dropping close to zero beyond age 80. The second panel is labeled “Panel B: Distribution of the value of the investment portfolios in the sample in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Portfolio value” and shows six categories, marked from left to right as follows: 1-less than 10,000 P L N, 2-10,000 to 25,000 P L N, 3-25,000 to 50,000 P L N, 4-50,000 to 100,000 P L N, 5-100,000 to 500,000 P L N, and 6-greater than 600,000 P L N. The vertical axis is labeled “Number of observations” and ranges from 0 to 250 in increments of 50 units. The data presented in the left graph (2019) is as follows: 1-less than 10,000 P L N: 109.93 2-10,000 to 25,000 P L N: 135.62 3-25,000 to 50,000 P L N: 127.05 4-50,000 to 100,000 P L N: 157.02 5-100,000 to 500,000 P L N: 210.96 6-greater than 600,000 P L N: 49.14 The data presented in the right graph (2020) is as follows: 1-less than 10,000 P L N: 70.79 2-10,000 to 25,000 P L N: 93.13 3-25,000 to 50,000 P L N: 121.48 4-50,000 to 100,000 P L N: 147.25 5-100,000 to 500,000 P L N: 203.09 6-greater than 600,000 P L N: 87.11 The third panel is labeled “Panel C: The distribution of the number of stocks in investor's portfolios in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Number of stocks in portfolio” and ranges from 0 to 120 in increments of 20 units. The vertical axis is labeled “Number of observations” and ranges from 0 to 250 in increments of 50 units. The 2019 (left) graph shows a histogram with the highest bar at around 200 observations for portfolios containing about 3 to 5 stocks. The number of observations gradually decreases as the number of stocks increases, with fewer than 20 observations beyond 40 stocks and near zero by 100 stocks. The 2020 (right) graph shows a similar histogram, with the highest bar at around 170 observations for portfolios containing about 3 to 5 stocks. The number of observations also declines steadily with increasing stocks, showing fewer than 20 observations beyond 40 stocks and close to zero by 100 stocks. The fourth panel is labeled “Panel D: The share of savings invested in the stock market in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Share of savings invested in stocks” and shows five categories, marked from left to right as follows: 0–0 percent, 1–less than 25 percent, 2–25 to 50 percent, 3–50 to 75 percent, and 4–greater than 75 percent. The vertical axis is labeled “Number of observations” and ranges from 0 to 350 in increments of 50 units. The data presented in the left graph (2019) is as follows: 0–0 percent: 20.27 1–less than 25 percent: 303.33 2–25 to 50 percent: 228.74 3–50 to 75 percent: 177.1 4–greater than 75 percent: 113.99 The data presented in the right graph (2020) is as follows: 0–0 percent: 14.62 1–less than 25 percent: 256.92 2–25 to 50 percent: 193.46 3–50 to 75 percent: 178.08 4–greater than 75 percent: 126.15 The fifth panel is labeled “Panel E: Assessment of investor’s skills by respondents in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Investors skills” and shows five categories, marked from left to right as follows: 1, 2, 3, 4, and 5. The horizontal axis is also marked “Low” on the left and “High” on the right. The vertical axis is labeled “Number of observations” and ranges from 0 to 450 in increments of 50 units. The data presented in the left graph (2019) is as follows: 1: 78.13 2: 196.81 3: 377.31 4: 88.02 5: 16.32 The data presented in the right graph (2020) is as follows: 1: 63.3 2: 224.01 3: 397.09 4: 73.19 5: 6.43 The sixth panel is labeled “Panel F: The frequency of portfolio rebalancing in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Portfolio rebalancing frequency” and ranges from 0 to 3 in increments of 0.5 units and is also marked with six categories, from left to right as follows: Irregularly, Quarterly, Monthly, Once a week, With high variability, and Daily. The vertical axis is labeled “Number of observations” and ranges from 0 to 250 in increments of 50 units. The data presented in the left graph (2019) is as follows: Irregularly: 218.21 Quarterly: 90.49 Monthly: 121.74 Once a week: 0 With high variability: 165.22 Daily: 242.99 The data presented in the right graph (2020) is as follows: Irregularly: 193.44 Quarterly: 59.56 Monthly: 118.31 Once a week: 0 With high variability: 155.19 Daily: 243.99 Note: All numerical data values are approximated.: Investor profile 2019–2020. Source: Authors’ own elaboration
The figure shows six panels arranged in a vertical series with two charts per row, the left chart for 2019 and the right chart for 2020. The first panel is labeled “Panel A: Age of the investors participating in the survey in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Investors age” and ranges from 10 to 100 in increments of 10 units. The vertical axis is labeled “Number of observations” and ranges from 0 to 140 in increments of 20 units. The 2019 (left) graph shows a histogram with a low count around ages 10 to 20, rising sharply from 30 to about 100 observations at age 30. The distribution peaks at around 110 observations between ages 35 to 40, remains close to 100 observations for ages 40 to 50, then declines to about 70 observations at age 55. Counts drop to about 40 around age 60, then rise again to 60 to 70 observations between ages 65 to 70, and decline gradually to near zero by age 90. The 2020 (right) graph shows a similar histogram, with fewer than 20 observations around ages 10 to 20. The count increases steeply to around 70 observations at age 30, peaking at about 120 observations between ages 35 to 40, and remaining above 100 through ages 40 to 45. Counts gradually decrease to about 90 at age 50, 60 at age 55, and 30 to 40 between ages 60 to 70, before dropping close to zero beyond age 80. The second panel is labeled “Panel B: Distribution of the value of the investment portfolios in the sample in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Portfolio value” and shows six categories, marked from left to right as follows: 1-less than 10,000 P L N, 2-10,000 to 25,000 P L N, 3-25,000 to 50,000 P L N, 4-50,000 to 100,000 P L N, 5-100,000 to 500,000 P L N, and 6-greater than 600,000 P L N. The vertical axis is labeled “Number of observations” and ranges from 0 to 250 in increments of 50 units. The data presented in the left graph (2019) is as follows: 1-less than 10,000 P L N: 109.93 2-10,000 to 25,000 P L N: 135.62 3-25,000 to 50,000 P L N: 127.05 4-50,000 to 100,000 P L N: 157.02 5-100,000 to 500,000 P L N: 210.96 6-greater than 600,000 P L N: 49.14 The data presented in the right graph (2020) is as follows: 1-less than 10,000 P L N: 70.79 2-10,000 to 25,000 P L N: 93.13 3-25,000 to 50,000 P L N: 121.48 4-50,000 to 100,000 P L N: 147.25 5-100,000 to 500,000 P L N: 203.09 6-greater than 600,000 P L N: 87.11 The third panel is labeled “Panel C: The distribution of the number of stocks in investor's portfolios in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Number of stocks in portfolio” and ranges from 0 to 120 in increments of 20 units. The vertical axis is labeled “Number of observations” and ranges from 0 to 250 in increments of 50 units. The 2019 (left) graph shows a histogram with the highest bar at around 200 observations for portfolios containing about 3 to 5 stocks. The number of observations gradually decreases as the number of stocks increases, with fewer than 20 observations beyond 40 stocks and near zero by 100 stocks. The 2020 (right) graph shows a similar histogram, with the highest bar at around 170 observations for portfolios containing about 3 to 5 stocks. The number of observations also declines steadily with increasing stocks, showing fewer than 20 observations beyond 40 stocks and close to zero by 100 stocks. The fourth panel is labeled “Panel D: The share of savings invested in the stock market in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Share of savings invested in stocks” and shows five categories, marked from left to right as follows: 0–0 percent, 1–less than 25 percent, 2–25 to 50 percent, 3–50 to 75 percent, and 4–greater than 75 percent. The vertical axis is labeled “Number of observations” and ranges from 0 to 350 in increments of 50 units. The data presented in the left graph (2019) is as follows: 0–0 percent: 20.27 1–less than 25 percent: 303.33 2–25 to 50 percent: 228.74 3–50 to 75 percent: 177.1 4–greater than 75 percent: 113.99 The data presented in the right graph (2020) is as follows: 0–0 percent: 14.62 1–less than 25 percent: 256.92 2–25 to 50 percent: 193.46 3–50 to 75 percent: 178.08 4–greater than 75 percent: 126.15 The fifth panel is labeled “Panel E: Assessment of investor’s skills by respondents in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Investors skills” and shows five categories, marked from left to right as follows: 1, 2, 3, 4, and 5. The horizontal axis is also marked “Low” on the left and “High” on the right. The vertical axis is labeled “Number of observations” and ranges from 0 to 450 in increments of 50 units. The data presented in the left graph (2019) is as follows: 1: 78.13 2: 196.81 3: 377.31 4: 88.02 5: 16.32 The data presented in the right graph (2020) is as follows: 1: 63.3 2: 224.01 3: 397.09 4: 73.19 5: 6.43 The sixth panel is labeled “Panel F: The frequency of portfolio rebalancing in 2019 (left) and in 2020 (right).” The horizontal axis in both graphs of this panel is labeled “Portfolio rebalancing frequency” and ranges from 0 to 3 in increments of 0.5 units and is also marked with six categories, from left to right as follows: Irregularly, Quarterly, Monthly, Once a week, With high variability, and Daily. The vertical axis is labeled “Number of observations” and ranges from 0 to 250 in increments of 50 units. The data presented in the left graph (2019) is as follows: Irregularly: 218.21 Quarterly: 90.49 Monthly: 121.74 Once a week: 0 With high variability: 165.22 Daily: 242.99 The data presented in the right graph (2020) is as follows: Irregularly: 193.44 Quarterly: 59.56 Monthly: 118.31 Once a week: 0 With high variability: 155.19 Daily: 243.99 Note: All numerical data values are approximated.: Investor profile 2019–2020. Source: Authors’ own elaboration
The average value of an investor’s portfolio among respondents in 2019 was PLN 207.589 (median PLN 100.000), with men keeping slightly larger portfolios (PLN 211.389 on average) than women (PLN 172.558). Figure 2 (Panel B) presents the distribution of the value of portfolios in the sample according to the adopted scale (1 – portfolios up to PLN 10.000, 2 – up to PLN 25.000, 3 – up to PLN 50.000, 4 – up to PLN 100.000, 5 up to PLN 500.000, 6 – over PLN 650.000). The share of stock market investments in savings averaged 25–50% in the entire group, with a slightly higher share of investments in men’s portfolios (2% difference). Invested funds source was mainly their work, only 2.5% of investors invest funds from inheritance or winnings, which could be significant for risk aversion. During the pandemic, the average value of investors’ portfolios increased to PLN 251.570 (women – PLN 201.122, men – PLN 255.486). However, the share of stock exchange investments in savings (25–50%) and the main source of funds for investments (work) did not change, but we observed an increase in funds from an inheritance or a win (6.5%).
The investment portfolios consisted mainly of shares (95.5% of respondents declared investing in shares, in 2020 it was almost 98%), with the number of stocks in the portfolios averaging 8 (10.2 in 2020), and at the time of participation in the survey in 2019, nearly 8% of respondents had no shares in their portfolio. According to the research by Gluzicka (2016), which served as the only available benchmark for our results in this part, well-diversified portfolios accounted for less than 3% of the sample (2.85%). Figure 2 (Panel C) presents the distribution of the number of shares in investors’ portfolios.
The assets in investors’ portfolios also included, on average, 0.73 (1.97 in 2020) shares of companies listed on NewConnect, 1.17 (1.11) bonds, 0.51 (0.35) futures, 0.30 (0.31) alternative instruments, 0.52 (0.90) foreign shares, 0.4 (0.53) shares or stocks of unlisted companies and 1.63 (1.33) shares in investment funds. Noteworthy, in 2019, on average, the value of investor’s portfolios in the study accounted for 25–50% of savings (Figure 2 (Panel D); scale 0–0%, 1 – up to 25%, 2–2,550%, 3–50–75%, 4–75–100%).
The respondents in both studies had higher education (81% in 2020) and 5–10 years of market experience (median). The median assessment of their skills in both surveys was three out of five (Figure 2, Panel E).
An important part of the study is also to determine how respondents manage their portfolios. According to the results of both studies, respondents conducted portfolio rebalancing less than once a week, but more often than once a month (Figure 2, Panel F; scale: 0 – irregularly, 0.5 – quarterly, 1 – monthly, 2 – once a week, 2.5 – with high variability, 3 – daily). However, in this case, the distribution of responses is important, as it shows that investors mainly perform daily reconstruction (3) or do it irregularly (0). Furthermore, on average, investors conduct more than six transactions per month. In the case of tools used in the decision-making process, in 2019 (2020), 64% (57%) of respondents indicated technical analysis, 59% (31%) fundamental analysis and over 41% (45%) of investors relied on recommendations. Importantly, due to the subsequent analysis of the behavioral biases in decision-making, 32% (29%) of the respondents tracked the historical minimum and maximum values, and 51% (54%) relied on recent stock price rises and falls.
Research design
The main goal of the article was to verify the hypothesis that the COVID-19 pandemic significantly changed investors’ behavior in financial markets and then to investigate the changes in investors’ preferences after the COVID-19 outbreak.
To illustrate the pandemic’s impact on investor’s utility and expectations, we conducted the responses’ quantification. This included finding the risk and loss aversion parameter to find the changes to the shape of the value function from prospect theory as well as identifying the weights applied by the investors to the observed probabilities. The final factor to be identified was hyperbolic discounting which would reflect the changes in the market expectations. The overall process should help identify whether investors became more or less rational after the pandemic outbreak.
The utility function we used for the research was modified Lucas utility given by formula 1 (Lucas, 1978), where the first part of the right side is responsible for consumption and the second for the investments. Thus, the open question was whether to solve it assuming the dependence of the stock price and consumption or not (Barberis et al., 2001).
in which:
- investor utility,
- consumption,
- benchmark defined by the expected portfolio value, its current and past values,
- , where is benchmark rate of return and is a risk-free rate,
- current portfolio value,
- investor’s prior gains and losses,
- scaling factor,
- hyperbolic discounting factor,
- utility consumption parameter depending on consumption rate.
To calculate the risk aversion factor, we adopted the approach of Jordà, Schularick, and Taylor (2019) based on the assumption that there exists a dependency between gains on stocks and consumption for the investors. We calculated the loss aversion factor based on the answers given to questions analyzing the curvature (Appendix 1, questions – 12, 17, 18, 19) of the prospect’s theory value function (blue line in Figure 3) based on the reaction to gains and losses through the reference point and whether the answers given by the same respondent are consistent, same as Kahneman and Tversky (1979) did in their original work. The control questions included responses from areas covering the value of the portfolio, capital sources and stated loss tolerance. We based the weights applied to the observed probabilities on the analysis of the factors shaping investors’ decision-making process.
The figure shows six panels arranged in a vertical series with two charts per row, the left chart for 2019 and the right chart for 2020. The first panel is labeled “Panel A:The diversification of portfolios according to its value.” The horizontal axis in both scatter plots is labeled “Portfolio value” and is marked with six categories from left to right as follows: 1–less than 10,000 P L N, 2–10,000 to 25,000 P L N, 3–25,000 to 50,000 P L N, 4–50,000 to 100,000 P L N, 5–100,000 to 500,000 P L N, and 6–greater than 600,000 P L N. The vertical axis is labeled “Number of stocks in portfolio” and ranges from 0 to 100 in increments of 20 units. The left graph shows a line in an upward trend, which starts at the number of 5 stocks in category 1 and ends nearly at the number of 15 stocks in category 6. The right graph shows a line in an upward trend, which starts at the number of 6 stocks in category 1 and ends nearly at the number of 20 stocks in category 6. The second panel is labeled “Panel B: Portfolio value and investor’s age.” The horizontal axis in both scatter plots is labeled “Portfolio value” and is marked with six categories from left to right as follows: 1–less than 10,000 P L N, 2–10,000 to 25,000 P L N, 3–25,000 to 50,000 P L N, 4–50,000 to 100,000 P L N, 5–100,000 to 500,000 P L N, and 6–greater than 600,000 P L N. The vertical axis is labeled “Age of investors” and ranges from 0 to 100 in increments of 20 units. The left graph shows a line in a slight upward trend, which starts at the age of about 35 years in category 1 and ends at about 55 years in category 6. The right graph shows a line in a slight upward trend, which starts at the age of about 30 years in category 1 and ends at about 45 years in category 6. The third panel is labeled “Panel C: Investor’s experience and the number of stocks in the portfolio.” The horizontal axis in both scatter plots is labeled “Investors skills” and shows five categories, marked from left to right as follows: 1, 2, 3, 4, and 5. The horizontal axis is also marked “Low” on the left and “High” on the right. The vertical axis is labeled “Number of stocks in portfolio” and ranges from 0 to 100 in increments of 20 units. The left graph shows a nearly flat line, which starts at about 5 stocks in category 1, remains almost constant across categories 2 to 4, and ends at about 7 stocks in category 5. The right graph shows a similar, nearly flat line, which starts at about 7 stocks in category 1, slightly rises through categories 2 and 3, and ends at about 8 stocks in category 5. The fourth panel is labeled “Panel D: The number of stocks in the portfolio and the probability of a crisis.” The horizontal axis in both scatter plots is labeled “Number of stocks in portfolio” and ranges from 0 to 100 in increments of 20 units. The vertical axis is labeled “Crisis expectations” and ranges from 0 to 10 in increments of 2 units. The left graph shows scattered points concentrated between 0 and 40 stocks, spread across all crisis expectation levels from 0 to 10. A red line runs vertically, fluctuating slightly around 0 to 10 crisis expectations while remaining close to the lower stock count values. The right graph also shows scattered points concentrated between 0 and 40 stocks, distributed across all levels of crisis expectations from 0 to 10. A red line again runs vertically, shifting slightly but staying near the lower stock count range. Note: All numerical data values are approximated.Investors’ characteristics. Source: Authors’ own elaboration
The figure shows six panels arranged in a vertical series with two charts per row, the left chart for 2019 and the right chart for 2020. The first panel is labeled “Panel A:The diversification of portfolios according to its value.” The horizontal axis in both scatter plots is labeled “Portfolio value” and is marked with six categories from left to right as follows: 1–less than 10,000 P L N, 2–10,000 to 25,000 P L N, 3–25,000 to 50,000 P L N, 4–50,000 to 100,000 P L N, 5–100,000 to 500,000 P L N, and 6–greater than 600,000 P L N. The vertical axis is labeled “Number of stocks in portfolio” and ranges from 0 to 100 in increments of 20 units. The left graph shows a line in an upward trend, which starts at the number of 5 stocks in category 1 and ends nearly at the number of 15 stocks in category 6. The right graph shows a line in an upward trend, which starts at the number of 6 stocks in category 1 and ends nearly at the number of 20 stocks in category 6. The second panel is labeled “Panel B: Portfolio value and investor’s age.” The horizontal axis in both scatter plots is labeled “Portfolio value” and is marked with six categories from left to right as follows: 1–less than 10,000 P L N, 2–10,000 to 25,000 P L N, 3–25,000 to 50,000 P L N, 4–50,000 to 100,000 P L N, 5–100,000 to 500,000 P L N, and 6–greater than 600,000 P L N. The vertical axis is labeled “Age of investors” and ranges from 0 to 100 in increments of 20 units. The left graph shows a line in a slight upward trend, which starts at the age of about 35 years in category 1 and ends at about 55 years in category 6. The right graph shows a line in a slight upward trend, which starts at the age of about 30 years in category 1 and ends at about 45 years in category 6. The third panel is labeled “Panel C: Investor’s experience and the number of stocks in the portfolio.” The horizontal axis in both scatter plots is labeled “Investors skills” and shows five categories, marked from left to right as follows: 1, 2, 3, 4, and 5. The horizontal axis is also marked “Low” on the left and “High” on the right. The vertical axis is labeled “Number of stocks in portfolio” and ranges from 0 to 100 in increments of 20 units. The left graph shows a nearly flat line, which starts at about 5 stocks in category 1, remains almost constant across categories 2 to 4, and ends at about 7 stocks in category 5. The right graph shows a similar, nearly flat line, which starts at about 7 stocks in category 1, slightly rises through categories 2 and 3, and ends at about 8 stocks in category 5. The fourth panel is labeled “Panel D: The number of stocks in the portfolio and the probability of a crisis.” The horizontal axis in both scatter plots is labeled “Number of stocks in portfolio” and ranges from 0 to 100 in increments of 20 units. The vertical axis is labeled “Crisis expectations” and ranges from 0 to 10 in increments of 2 units. The left graph shows scattered points concentrated between 0 and 40 stocks, spread across all crisis expectation levels from 0 to 10. A red line runs vertically, fluctuating slightly around 0 to 10 crisis expectations while remaining close to the lower stock count values. The right graph also shows scattered points concentrated between 0 and 40 stocks, distributed across all levels of crisis expectations from 0 to 10. A red line again runs vertically, shifting slightly but staying near the lower stock count range. Note: All numerical data values are approximated.Investors’ characteristics. Source: Authors’ own elaboration
We estimated the last parameter, hyperbolic discounting, from the responses to the questions concerning expectations (Appendix 1, questions – 15 and 16). The control questions were in that case the ones on expectation and probability of market crash, the ratio of savings to investments, the origin of the invested capital and the impact of investment results on consumption/everyday life.
Finally, we fitted the logistic function to the dataset consisting of responses to questions concerning risk and loss aversion. This approximation served as an investor’s utility function (value function from the prospect theory) based on the findings from our surveys for 2019 and 2020.
Results and discussion
We analyzed the regularities and differences in investor profiles presented in the previous section. We also confronted them with research on the psychological aspects of decision-making. We tested the presented differences for statistical significance using the t-test (for the significance level 1–10%).
Changes in the portfolio construction before and after COVID-19
The first characteristic analyzed was the degree of portfolio diversification depending on its value. The relationship measured by Rho-spearman’s correlation was positive and characterized by a coefficient of 0.33 in 2020 and 0.34 in 2019. Although the correlation remained at a similar level in both periods, the number of shares in the portfolio increased during the pandemic (as well as the portfolio’s value among the sample). On average, the number of stocks ranged from 3.8 for the smallest portfolios to 12.5 for the biggest ones in 2019, while it shifted to 4.0 to 14.8 respectively in 2020 (the mean for each portfolio value in 2019 and 2020 was marked by the red line on Figure 3, Panel A). We may assume that the higher number of shares in an average investor’s portfolio resulted from a greater need for diversification due to greater volatility and greater uncertainty in the financial markets during the pandemic.
Figure 3 (Panel B) presents the portfolio values against the investors’ age. Although there was a positive relationship between investor age and portfolio values (Table 1), increasing age was mainly associated with increases in the minimum and average (red line) portfolio values. However, we did not confirm a similar relationship in the case of the top portfolio value. The average age of the investors holding the smallest portfolios slightly fell to 34.3 in 2020 (34.4 in 2019). For the largest portfolios, the average age fell from 50.1 in 2019 to 49.3 in 2020 (p-value<0,1). Age might relate to the propensity to take risks (Bellante & Green, 2004). During the pandemic, many new investors entered the market, tempted by the high volatility. This may explain the decline in the average age of investors during the period of high market volatility. Research has shown that younger investors were more likely to use technical analysis tools than fundamental analysis. This may result from a shorter investment horizon and less affluent investment portfolios.
Correlations comparison between investment aspects in 2019 and 2020
| 2020 | t-statistics | Significance | 2019 | t-statistics | Significance | |
|---|---|---|---|---|---|---|
| Portfolio value – stocks | 0.33128 | 9.71118 | *** | 0.34064 | 10.49441 | *** |
| Portfolio value – age | 0.31311 | 9.11857 | *** | 0.30727 | 9.35257 | *** |
| Experience – stocks | 0.11658 | 3.24672 | *** | 0.16792 | 4.93393 | *** |
| Crisis probability – stocks | −0.07399 | −2.05195 | ** | −0.02762 | −0.80029 |
Note(s): Significance level: *** − 0.01; ** − 0.05; * − 0.1
As expected, the investors’ responses showed a clear correlation between their experience and the number of stocks in the portfolios (Figure 3, Panel C). However, during the pandemic, the number of shares in the portfolio increased in all of the portfolios, but more in the portfolios of less experienced investors. This may prove the awareness of the increased investment risk among investors. The average number of stocks for the less experienced investors shifted from 4.6 in 2019 to 7.7 in 2020, while for the most experienced ones it changed from 9.5 to 11.5. At the same time, the correlation between experience and the number of stocks decreased from 0.168 to 0.117 (Table 1). This may indicate an influx of new, relatively inexperienced investors into the market during the pandemic lock-up and reflect this situation in our sample.
The events of 2020 led to an analysis of the likelihood of a financial crisis understood as a market collapse expected by investors and its impact on the actions taken when the composition of the portfolio is considered. The median of crisis expectations was 5 (out of 10) both in 2019 and 2020. This may come as a surprise, but it should be borne in mind that we conducted the 2020 survey after the market had recovered after the stock market crash in the first months of 2020. After this period, the market grew dynamically, gradually making up for losses, which seems like the investor’s reaction to the initial overreaction to the pandemic outbreak. However, there may be some changes in the approach to portfolio construction as a result of the experience from the beginning of the pandemic. According to the data presented in Figure 3 (Panel D), the higher the expected probability of a crisis occurrence, the fewer stocks are held in the portfolio. While this is true for both 2019 and 2020, investors expecting the crisis (10 on the scale) in 2020 have become relatively more cautious compared to the group of other investors. A much higher level of risk in the market led to a drop in the number of stocks in this group from 7.7 to 6.9. This was likely due to the reduction in their risk exposure rather than a reduction in the level of diversification. Those who expect a very low probability of a crisis (1 on the scale) tend to invest and diversify their portfolios more – on average 13.5 in 2020 compared to 9.1 in 2019. That may be the result of investors exploiting the situation of high volatility on the market and possible gains in 2020, as they expect the probability of a crisis as slightly higher. However, the low relationship between the perception of the likelihood of a crisis and the number of shares in the portfolio in 2019 (−0.027) was not statistically significant (Table 1).
Impact of COVID-19 on decision-making process
In terms of the data and analysis presented, it is important to assess the impact of the market crash in March 2020 on investors’ preferences. Comparing the results of the 2019 and 2020 surveys allows for an in-depth analysis of both investors’ expectations and changes in their utility functions. The starting point for utility is the value function from the prospect theory (Formula 1).
Concerning preferences, there was a clear impact of the pandemic on portfolios’ diversification (an increase in the number of stocks from 8 to more than 10 in 2020) and increased exposure to companies listed on the NewConnect market (an increase from 0.73 in 2019 to 1.97 in 2020), foreign stocks (from 0.52 to 0.90), shares or stocks of non-listed companies (from 0.4 to 0.53) and a reduction in the share of relatively safe investments in the portfolios – investment funds (a decrease from an average of 1.63 to 1.33). During the pandemic outbreak, the interest in alternative investments also increased among investors – in gold from 14% to 22%, coins from 10% to 12%, cryptocurrencies from 8% to 11% and startups and crowdfunding involvement from 6.5% to 13% as well as alcohol investments (from 2.3% to 2.9%). Simultaneously, the interest in investing in art decreased from 4% to 2.7%. One of the reasons for the increasing level of diversification may be the growth in declared expectations of the financial crisis in the 2020 survey (from 85% to 93%), while the probability of the crisis itself increased, according to the 2020 survey, by only 2% (from 51% to 53%). Moreover, the average number of transactions per month almost doubled (from 6 to 13).
Due to the research hypothesis formulated in the study, the most important conclusion seems to relate to three behavioral aspects, namely risk aversion, loss aversion and hyperbolic discounting. These three aspects are the result of both the investor’s expectations and the effect of experiences that influence the decisions made. While the analyses presented thus far have led to a higher frequency of transactions, a better diversification of portfolios and reaching for a greater number of alternative investments, it is interesting how it translates into the investor’s utility function.
Our research showed that risk aversion increased during the pandemic to 2.576 (from 2.088 in 2019, p-value<0.01), which makes it consistent with the findings of Vasileiou (2020) and Heo, Grable, and Rabbani (2021). Although this factor has an economic interpretation up to the level of 10 units (Arrow, 1971; Mehra & Prescott, 1985), most studies in the field of classical finance define it at a level close to 1 for rational investors (Arrow, 1971). We could explain the level of risk aversion, which was twice as high as the entry-level for a rational investor in 2019, by the fact that the investor required a risk premium twice as high as that of a rational investor according to Von Neumann and Morgenstern (1944). The nearly 25% increase in risk aversion in 2020 should be reflected in an increased number of stocks in the portfolios and the frequency of transactions (almost twofold increase in the number of transactions per month). Kaplow (2005) discussed further findings on relative risk aversion.
Moreover, the value of a loss aversion (λ), dropped to 0.82 in the 2020 study (from 1.06 in 2019). This is something of a contradiction to the assumptions of the prospect theory, in which Kahnemann and Tversky stated that the value of this indicator should be higher than 1. Its reduction below 1 means that there is indeed a different perception of profit and loss that actually exists among investors, but during the pandemic, it is in favor of profitable positions, which are held in portfolios for longer periods. This finding confirms Ruggeri et al. (2020) results on prospect theory foundations. Therefore, the consideration of a potential market crash discussed earlier affects the loss reduction (faster), even faster than in the case of neoclassical finance assumptions. This “mechanism” results from a possible quick make-up for losses on the volatile market during its recovery, when there are many stocks available as an alternative to the loss-making ones. Figure 4 shows the investor’s utility function and its shifts based on our research. In the experimental work of Walasek and Stewart (2015), and the meta-analysis of these factor values in Walasek et al. (2024), we may find further discussion on the loss aversion parameter, also below 1.
The vertical axis is labeled “Utility value,” and the horizontal axis is divided into two regions labeled “Losses” on the left and “Gains” on the right. Three curves are plotted on the graph: a blue line labeled “Prospect theory,” an orange line labeled “2019,” and a yellow line labeled “2020.” All three lines start from the losses side at the lower left, curve upward sharply near the origin, and flatten at different levels toward the gains side on the right.Investor’s utility function (value function from the prospect theory: blue line). Source: Authors’ own elaboration
The vertical axis is labeled “Utility value,” and the horizontal axis is divided into two regions labeled “Losses” on the left and “Gains” on the right. Three curves are plotted on the graph: a blue line labeled “Prospect theory,” an orange line labeled “2019,” and a yellow line labeled “2020.” All three lines start from the losses side at the lower left, curve upward sharply near the origin, and flatten at different levels toward the gains side on the right.Investor’s utility function (value function from the prospect theory: blue line). Source: Authors’ own elaboration
The changes in the utility function observed in Figure 4 may potentially have a huge impact on the portfolios, pricing and market as a whole. Therefore, it is important to indicate that observed changes in the utility result from the biases discussed in the article, such as market crash expectations, the disposition effect and observed historical maximum or minimum prices, to name a few.
These factors, analyzed together, illustrate well the impact of the March 2020 market crash on investor preferences – an increase in risk aversion leads to a faster closing of losing and winning positions, but the decrease in loss aversion means that investors would not be interested in maintaining lossy positions longer than profitable ones. This is consistent with the Gupta and Shrivastava (2022) findings justified by the fear of missing out (FOMO). The observed inversion of the utility function in the lossy part (3rd quarter of the coordinate plane) suggests that extreme shifts in market volatility can supposedly cause investors to be more rational than prospect theory suggests. Furthermore, it can be the result of investors becoming more rational as well as opportunity-driven (greedy), as the market rebound after the pandemic outbreak was relatively sharp. However, based on our data, it can be all factors together.
The last factor analyzed in this section is hyperbolic discounting, which fell from 6% in 2019 to 4.6% in 2020, implying that investors have lowered their expectations of deferred benefits. Together with the changes in utility discussed earlier in the text, this may indicate that greater uncertainty prompts investors to take profits faster, and in the case of calculating the expected value, the lower probability of receiving an additional profit has a greater impact than the value of the potential payout alone. This finding fills a gap in the literature on investor behavioral biases, as it is not widely identified, which could be due to a lack of granular data.
Conclusions
According to our study, the investor on the WSE is usually a middle-aged man with a relatively large and relatively well-diversified portfolio (over PLN 200,000, shares in eight companies in 2019 while over PLN 250,000 and over 10 companies in the portfolio in 2020). When investing, the WSE investors rely on technical analysis, followed by fundamental one and market recommendations. They also monitor and rebalance their portfolios relatively frequently.
The most important finding of our research is the identified shift in risk aversion, change in loss aversion and hyperbolic discounting. However, the changes in the investor’s utility function based on COVID-19 are likely to be the result of many biases combined, reflecting changes in the investor’s attitude towards the market situation and, in particular, market risk. This may lead to a more rational approach to investing, especially when losses are concerned, and may therefore lead to changes in the anomalies discussed in behavioral finance, but as such requires further analysis. More importantly, our study challenged the foundations of prospect theory, where the loss aversion parameter is less than one. This type of argument is not new, as it has been mentioned in many articles such as List (2004), Maialeh (2019) and Ruggeri et al. (2020). The challenge to prospect theory in favor of neoclassical theory is based solely on the individual investors from Poland during the very unusual period and requires further investigation. Moreover, the notion that investors behave more rationally, especially during market turbulences, may lead to greater market efficiency during such periods. This is an interesting, yet obvious finding, that market turbulence leads to the reduction in pricing inefficiencies. Noteworthy, our perspective supports it on the investor-level characteristics.
Consequently, our respondents tended to be more risk-averse, and on average they had more assets in their portfolios during periods of shifted volatility. This seems very rational and there are two possible explanations for this state of affairs. Either they aimed to maintain more diversified portfolios to limit risk, or, given the near-zero interest rates and high market volatility, they did not hold capital in deposits while taking advantage of the ups and downs of the market. Moreover, the higher frequency of trading observed among investors may have a significant impact on the market as a whole in the event of anomalies, as it may limit the growth of bubbles and amplify market reactions – again in favor of the neoclassical efficient market hypothesis rather than behavioral economics. This can potentially minimize the impact of market microstructure frictions on market prices. This way, our research proves that there are visible effects at the market level during periods of increased volatility due to factors stemming from the investors’ characteristics.
The supplementary material for this article can be found online.
